High performance computing (HPC) is a core component of the Center for Nutritional Immunology and Molecular Medicine Laboratory (NIMML). HPC runs in silico experiments for generating novel hypotheses and processes the data generated from the high-throughput lab (HTL).
The MIEP program of NIMML focuses on modeling and simulation of immune responses for enteric pathogens and runs large numbers of in silico experiments. Due to the high complexity of the immune responses, modeling these processes and running in silico experiments depends highly on availability of robust computing resources. For example, a COPASI CD4+ T cell differentiation model can take more than one week to fully estimate parameters. In addition, an ENISI in silico experiment with more than 1 million cells could take more than 20 hours to run. The hypotheses generated from the in silico experiments will be validated through biological experiments in the laboratory. The high-throughput data obtained from the biological experiments requires HPC for processing, analyses and visualization. Ultimately this integrated effort will help identify potential therapeutic targets and novel disease biomarkers.
Current resources at the Core Computational Facility (CCF) at the Virginia Bioinformatics Institute (VBI) encompass more than 7.4 TB of RAM, 1860 processor cores distributed across 250 servers and clusters, and close to 4.3 PB of storage, including 20 nodes in the cluster purchased by the MIEP program, with a total of 240 cores and 960 GB of main memory. The centers host a storage area network ensuring high-speed data access and reliability. The largest system is Shadowfax, which is optimized for HPC in the life and medical sciences, a hybrid cluster with 912 processor cores, 5.4 TB of RAM, 40Gb/s InfiniBand network, 80TB parallel storage, 16 nVidia Tesla GPGPUs (7168 CUDA cores) and 3 FPGA based Convey HC-1 systems. Off-site data backup and recovery procedures prevent data loss in case of a disaster. The CCF provides the technical infrastructure for experimental data storage (high-throughput reporter assays, flow cytometry, transcriptomics sequencing, proteomics, ChipSeq).
The live stats for resource usages of our data center can be monitored here.
